Scribd
Upload
59 views

Experimental Methods: Department of Applied Mechanics

The document discusses frequency distributions of rainfall data from Delhi and Bombay over many years. It shows that Bombay receives more rainfall on average than Delhi. Histograms are used to compare the distributions, showing that the most likely number of rainy days in Bombay is between 79-80 days, while in Delhi it is between 28-29 days. The document also discusses how continuous frequency distributions can be approximated from discrete data by treating intervals as discrete events.

Uploaded by

ANSHU MISHRA
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
59 views

Experimental Methods: Department of Applied Mechanics

The document discusses frequency distributions of rainfall data from Delhi and Bombay over many years. It shows that Bombay receives more rainfall on average than Delhi. Histograms are used to compare the distributions, showing that the most likely number of rainy days in Bombay is between 79-80 days, while in Delhi it is between 28-29 days. The document also discusses how continuous frequency distributions can be approximated from discrete data by treating intervals as discrete events.

Uploaded by

ANSHU MISHRA
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 12

Department of Applied Mechanics

APL 103

Experimental Methods
Semester I, 2020-21

L2

Murali R Cholemari
Experimental Analysis
• How many days does it rain in Delhi?
• Some years it rains on more no. of days, some years it rains on less no. of
days.
• Say, we record the no. of days in the year on which it rains – say over hundred
and fifteen years. No. of days of rain xi No. of years ni
<25 3
25 4
26 5
27 10
28 20
29 32
30 22
31 11
32 5 How do we understand
>32 3 this??
Pictorial representation (histogram)
35 Frequency
30 Distribution
Delhi
25
frequency
No. of years

20

15

10

0
25 26 27 28 29 30 31 32 33
No. of Days

How many days it is most likely to rain next year??


• How many days does it rain in Bombay?
• It definitely rains more!
• The no. of days in the year on which it rains – say over hundred and fifty
years.
No. of days of rain xi No. of years ni
<75 9
75 9
76 10
77 15
78 25
79 42
80 27
81 16
82 10 How do we understand
>82 7 this??
Histogram
45 Frequency
40
Distribution
35
Bombay
30
No. of years

25

20

15

10

0
75 76 77 78 79 80 81 82 82
No. of Days

How many days it is most likely to rain next year??


Comparison?
• How do we compare the two cities?
• Observation: It definitely rains more in Bombay. But,
• On an average how many days more?
• What is the most no. of days likely?
• Least no. of days likely?
• We saw frequency distributions, which are pictorial representations
of the frequency of occurrence fi vs the event values xi
• Let the total number of events (note: discrete events!) σ𝑖 𝑛𝑖 = 𝑁
• Define relative frequency: 𝑓𝑖 = 𝑛𝑖/𝑁
Relative frequency distribution fi vs xi

0.3
0.3
0.25
0.25
0.2 Delhi 0.2 Bombay
0.15 0.15
fi

fi
0.1 0.1
0.05
0.05
0
25 26 27 28 29 30 31 32 33 0
75 76 77 78 79 80 81 82 82
No. of Days
No. of Days

Note that σfi= 1 . The frequency distribution is normalized


Limiting frequency distribution
• As a limiting distribution of a very large number of data points –
ideally infinite number.
• Fluctuations between distributions with smaller sized samples are
smoothed out.
• Represents the underlying physical phenomenon.
• Is an idealization (for the ‘infinite experiment’)
• Frequencies represented without subscript: 𝑓 𝑥𝑖
Continuous variables?
• E.g. Amount of rain on a given day?
• > define intervals and count the occurrences within these intervals —this
is now discrete
• Let 𝑥 be continuous, choose intervals 𝑥0 , 𝑥1 … . . 𝑥𝑚 . If frequencies within
these intervals are 𝑛1 , 𝑛2 , 𝑛3 … 𝑛𝑚 , then the relative frequencies are (n is
the total number of observations):
• 𝑛1 Τ𝑛 with 𝑥0 ≤ 𝑥 ≤ 𝑥1 ,
• 𝑛2 Τ𝑛 with 𝑥1 ≤ 𝑥 ≤ 𝑥2 ,

• 𝑛𝑚 Τ𝑛 with 𝑥𝑚−1 ≤ 𝑥 ≤ 𝑥𝑚 .
• Note: σ𝑖 𝑛𝑖 Τ𝑛 = 1 , the definition is normalized.
Continuous variables
• The heights of the histogram 𝑓𝑛 𝑥1 , 𝑓𝑛 𝑥2 , ⋯ 𝑓𝑛 𝑥𝑚 such that
• 𝑥1 − 𝑥0 𝑓𝑛 𝑥1 = 𝑛1 Τ𝑛 is the relative frequency in 𝑥0 ≤ 𝑥 ≤ 𝑥1
• 𝑥2 − 𝑥1 𝑓𝑛 𝑥2 = 𝑛2 Τ𝑛 is the relative frequency in 𝑥1 ≤ 𝑥 ≤ 𝑥2

• 𝑥𝑚 − 𝑥𝑚−1 𝑓𝑛 𝑥𝑚 = 𝑛𝑚 Τ𝑛 is the relative frequency in 𝑥m−1 ≤ 𝑥 ≤ 𝑥m

• Area under the histogram


𝑛𝑖
• σ𝑖 𝑥𝑖 − 𝑥𝑖−1 𝑓𝑛 𝑥𝑖 = ෍ = 1 --- Normalized!
𝑖 𝑛

• Note that the intervals 𝑥1 − 𝑥0 , 𝑥2 − 𝑥1 … need not be equal.


• Note also: 𝑓𝑛 𝑥𝑖 is the relative frequency per unit interval – the actual relative
frequency in the ith interval is 𝑥𝑖 − 𝑥𝑖−1 𝑓𝑛 𝑥𝑖 .
Continuous Distribution
• How do we relate the two histograms? One for the discrete variable
and one for the continuous variable?
• Consider a smooth curve through a discrete value distribution

𝐹𝑛 𝑥 𝐹𝑛 𝑥1 = 𝑓𝑛 𝑥1 , 𝐹𝑛 𝑥2 = 𝑓𝑛 𝑥2 ⋯

𝑥1 𝑥2 𝑥𝑚
Histogram of a continuous variable
• 𝐹𝑛 𝑥 approximately equal to the height of
rectangle at the midpoint of the interval
• Corresponding areas are equal:
𝑥𝑖
‫ 𝑥 𝑛𝐹 𝑥׬‬ⅆ𝑥 = 𝑥𝑖 − 𝑥𝑖−1 𝑓𝑛 𝑥𝑖
𝑖−1
• Hence,
∞
𝐹𝑛 𝑥 • ‫׬‬−∞ 𝐹𝑛 𝑥 ⅆ𝑥=σ𝑖 𝑥𝑖 − 𝑥𝑖−1 𝑓𝑛 𝑥𝑖
𝑛𝑖
=෍ =1
𝑛
𝑖
• 𝐹𝑛 𝑥 is called the continuous frequency
distribution curve

You might also like